Stich, Stephen P.

Introduction: The Idea of Innateness

Readers are reminded that copyright subsists in this extract and the work from which it was taken. Except as
provided for by the terms of a rightsholder's licence or copyright law, no further copying, storage or distribution is
permitted without the consent of the copyright holder.

The author (or authors) of the Literary Work or Works contained within the Licensed Material is or are the author(s)
and may have moral rights in the work. The Licensee shall not cause or permit the distortion, mutilation or other
modification of, or other derogatory treatment of, the work which would be prejudicial to the honour or reputation of
the author.

Stich, Stephen P., (1975) "Introduction: The Idea of Innateness", Stich, Stephen P. (ed.), Innate ideas, 1-22,
University of California Press This is a digital version of copyright material made under licence from the
rightsholder, and its accuracy cannot be guaranteed. Please refer to the original published edition.

Every effort has been made to trace the copyright owner(s) of this material, and anyone claiming copyright should
get in touch with Heron at heroncopyright@publishingtechnology.com.

Licensed for use at the University of Sheffield for the course: "phi332 - Philosophy of Psychology" during the period
09/02/15 to 07/02/16.

Permission reference: 0520029615(1-22)98437

ISN: 0520029615
 Introduction:
The Idea of Innateness

Philosophical controversies are notoriously long-lived. And in

point of venerability the controversy around innate ideas and in-
nate knowlegdge is equal to any. It differs, however, from many
of its cousins of comparable ancestry. For in the last decade it
has emerged anew as a lively debate whose participants include
some of the most important philosophers in the English-speak-
ing world. The debate is unique, too, in having been rekindled
not by philosophers but by linguists who based their arguments
on the findings of modern generative grammar.
It is not surprising that in a controversy extending over
two millennia the strands of the argument have become knotted
and intertwined. My aim in this introductory essay is to un-
tangle a few of the strands of the argument, with the hope of
making it a little easier for the reader new to the debate to find
his bearings.
The controversy is easy enough to summarize: Some phi-
losophers, as well as linguists, psychologists, and others, allege
that human beings have innate knowledge or innate ideas. Others
deny it. But what is it to have innate knowledge or an innate
idea? There is a pattern running through much of the debate in
this area. Advocates of the doctrines of innate ideas and in-
nate knowlege commonly take the notion of innateness itself to
be unproblematic. They explain it with a few near synonyms,
inborn or unlearned, or with a metaphor or an allegory, and
leave it at that. The doctrines opponents often begin by puz-
zling over just what the doctrine could possibly mean. They go
on to construct a variety of accounts, arguing against each in
turn. The advocates rejoinder, as often as not, is that he has
been misunderstood. Thus, in approaching the debate over the
innateness doctrine, we would do well to ponder what we are
saying of someone when we say that he knows something or has
an idea innately.
In working toward an analysis of innateness there are two
2 INTRODUCTION

pitfalls we must avoid. First, we should unpack the concepts

in such a way that there might be innate ideas or innate know-
ledge. An account of these notions, which makes the claim
that a person has innate ideas or knowledge either straight-
forwardly logically impossible or patently empirically false,
holds little promise as an explication of what those who advance
the claim have in mind. Their view may be false but, if we
are to interpret them sympathetically, it will not be trivially
false. Second, our account should portray the innateness
doctrine as an interesting view about human cognitive mechan-
isms. An analysis would be suspect in this quarter if it entailed
that all knowledge or ideas are innate. Advocates of innateness
usually took themselves to be advancing an exciting thesis
about a special sort of knowledge or idea. But if, on a given
analysis, all knowledge or ideas are innate, there is reason to
be suspicious that this exciting thesis has been exchanged for
the humdrum claim that, on a special sense of innate, all
knowledge or ideas count as innate.
While counting all knowledge as innate is a symptom that
an account is philosophically uninteresting, it is not a sufficient
condition. In particular, we should note two historical caveats.
In some of his writings Plato seems to endorse the view that
all knowledge is innate. But on Platos view, only part of what
we commonly think we know is known innately. The rest
isnt worthy of being called knowledge at all. Platos move is
not to bloat the concept of innateness to encompass all
knowledge, but rather to shrink the concept of knowledge
until it coincides with what we know innately. Descartes, too,
sometimes maintains that all ideas are innate. However, as
Prof. Adams essay in this volume points out, there is both a
narrow and a broad sense in which an idea might be innate for
Descartes. When Descartes claims that all ideas are innate he
is using the broad sense. But the interesting hypothesis about
our cognitive mechanisms is the claim that we have innate
ideas in the narrow sense. While Descartes advocates this
latter hypothesis, he abjures the stronger one that all ideas are
innate in the narrow sense.
 INTRODUCTION 3

Innate Diseases
I observed . . . that there were in myself certain thoughts that did not
proceed from external objects, nor from a determination of my will, but
only from the thinking faculty that is in me; and therefore, in order to
distinguish the ideas or notions that are the content of these thoughts
from other ideas which are adventitious or manufactured, I called them
innate. It is in the same sense of the word that we say generosity is innate
in certain families; or again that in others certain diseases, e.g. gout and
the stone, are innate; not that infants of these families suffer from these
diseases in their mothers womb, but because they are born with a cer-
tain disposition or liability to contract them. (Descartes, Notes on a
Certain Programme)
In calling ideas innate, Descartes tells us, he is using the
same sense of the word we use when we say certain diseases are
innate. So let us launch our analysis of innateness by pursuing
Descartes hint and asking what it is to be afflicted with an
innate disease. Our strategy then will be to seek an analysis
of the notion of an innate disease. Armed with our analysis we
will return to tackle the thornier problem of innate knowledge.
To begin let us imagine a disease that at a certain stage,
is always characterized by a unique and easily observable set
of symptoms. (The lurid details are left to the reader.) In
imagining a disease always characterized by a unique set of
symptoms, we are making a simplifying assumption about the
relation between a disease and its symptoms. But the prey we
are stalking is innateness, not disease. A more realistic assump-
tion would complicate the discussion while shedding no further
light on the concept that interests us.
Now, under what conditions would we be willing to say
that someone having such a disease has it innately? A natural
first move is suggested by the parsing of innate as congeni-
tal or inborn. Perhaps to have the disease innately is just to
have the symptoms of the disease from birth. But, as Descartes
notes, this will not do. For a person may well have an in-
nate disease though none of its symptoms are evident at birth.
It may be that the symptoms appear only at some specific
stage later in lifeduring a certain age span, say, or accom-
panying some normal bodily change like puberty or menopause.
In such a case we are prepared to say that the person has the
4 INTRODUCTION

disease even before the appearance of the symptoms. Of course,

unless there is some way to predict the future occurrence of
the symptoms, we may not know the person has the disease
until he begins to exhibit the symptoms. Still, there is nothing
unusual about the claim that he had the disease all along,
though we didnt know it until the symptoms appeared. The
parallel to the notion of innate knowledge is clear. Those who
advocate the doctrine of innate knowledge are often willing
to attribute such knowledge to a person even though he has
not yet come to believe the proposition he is alleged to know.
But here we are getting ahead of ourselves.
We have, then, what appear to be two sorts of innate
infirmities, those whose symptoms are present at birth and
those whose symptoms appear only later. Let us focus for
a while on the second sort. Under what conditions can we
properly say that a person is afflicted with such an innate
malady? We have seen that the symptoms themselves need not
yet have appeared. So perhaps what is called for is an analysis
in the form of a conditional: To say that a person has a
disease of this second sort innately is to say that if he is of
the appropriate age (or at the appropriate stage of life) then
he has the symptoms.
To be at all plausible, this proposal demands at least one
modification. If the if . . . then locution it uses is understood
as a material conditional, true if the consequent is true or the
antecedent false, then the account as it stands has the conse-
quence that everyone who has yet to attain the appropriate
age has the disease innately. What is wanted, rather, is a
subjunctive locution forming not a material conditional but a
counterfactual conditional. With this modification, our analysis
becomes: To say that a person has a disease of the second
sort innately is to say that if he were of the appropriate age
(or at the appropriate stage of life) then he would have the
symptoms.
I think it could be argued that similar moves will be
needed in the analysis of the first sort of innate disease, the
cases where the symptoms are present at birth. This would,
for example, enable us to make sense of talk of a fetus having
an innate disease before showing any detectable abnormalities.
 INTRODUCTION 5

But rather than pursue this line, let us keep our attention
restricted to the second type of case. For ultimately the
subjunctive analysis will prove inadequate.
Returning, then, to cases of the second sort, let us
attend to a pair of further problems with the analysis as it
stands. First, consider the case of an infectious disease
caused, say, by a bacterial infection. Let us suppose that,
while the disease can be acquired at any age, the symptoms
appear during or after puberty. So a child may contract
the infection while still an infant. This, it would seem, is
a clear case of a person having a disease that is not innate.
Yet our analysis, as it stands, implies that the child has the
disease innately. From the time he contracts the infection
onward, it is true of him that if he were at puberty then he
would have the symptoms. Second, our analysis focuses on
the period of latency when the symptoms have yet to appear.
Thus it does not enable us to segregate innate from noninnate
diseases once the symptoms are present. Both these diffi-
culties can be patched if we swap a counterfactual locution
for an independent-of-factual conditional which is true of
the victim from the beginning of his life. We have, then: A
person has a disease innately if and only if, from the begin-
ning of his life it is true of him that if he is or were of the
appropriate age (or at the appropriate stage of life) then he
has or would have the diseases symptoms.
This leaves us with the nice problem of saying just when
a life begins. On the answer turns the distinction between
innate diseases and diseases caused by abnormal pregnancy.
Here I have no solutions to suggest. My suspicion is that the
distinction is a fuzzy one, and that on this score the notion
of innate disease is fuzzy. My only proposal is that for pur-
poses of our investigation we take life to begin sometime
before birth. This will collapse the distinction between the
two sorts of innateness, leaving our analysis applicable equally
to each.
Unhappily, our analysis is still not adequate. Its fault is
excessive pessimism. In defining the notion of innate disease
we have left no room for possible cures. Imagine an innate
disease whose symptoms, in the normal course of events,
6 INTRODUCTION

appear at age ten. Imagine further that a cure has been

developed and administered to one of the diseases victims at
age five. Intuitively, we want to say the young patient had
the disease until he was five. But this is blocked by our
account. For it is not true of him during his first five years
that if he were ten years old then he would have the symptoms.
Rather what is true of him during these years is a watered-
down subjunctive that we might render: If he were ten years
old, then, in the normal course of events, he would have the
symptoms. So our account, full-blown at last, becomes:
A person has a disease innately at time t if, and only if, from the
beginning of his life to t it has been true of him that if he is or were of
the appropriate age (or at the appropriate stage of life) then he has or
in the normal course of events would have the diseases symptoms.
Timid conditionals, hedged with in the normal course
of events and the like, are familiar to philosophers who have
reflected on the relation between counterfactual and dis-
positional locutions. Dispositionals, like x is soluble or x
is flexible, are commonly weaker than unhedged counter-
factuals.1 For to attribute a dispositional property to an
object is not to say what the object would do under certain
conditions, but rather to say what it would do under these
conditions if surrounding circumstances were normal or natural.
So we can, with some justice, dub our final analysis the
dispositional account of innateness. Descartes it seems had
much the same idea. Those who suffer innate diseases, on
his account, are born with a certain disposition or liability
to acquire them.
The notion of innate disease, if our dispositional account is
correct, is tied essentially to concepts of naturalness or normalcy.
The job of unpacking these concepts, of saying what is natural or
normal (or when other things are equal) is notoriously difficult.
While clear cases of normal situations and of abnormal ones
can be found, there is substantial vagueness in the middle. Our
analysis would lead us to expect that this vagueness is reflected in
the concept of innateness. The reflection is not hard to find.
Consider the distinction between an innate disease and
a susceptibility. To suffer from an innate disease is to be
disposed to acquire its symptoms at the characteristic time
 INTRODUCTION 7

in the normal course of events. To be susceptible to a (non-

innate) disease is to be disposed to acquire its symptoms
under certain special circumstances. Certain toxic diseases,
for example, can be acquired only by certain people. A
susceptible person, when exposed to the toxic substance, will
come down with the symptoms. At the extremes, the dis-
tinction seems clear enough. But notice how the two shade
into each other. Suppose a person becomes ill after ingesting
a certain amount of a particular chemical. (We can imagine
the effects to be cumulative.) Suppose also that the chemical
occurs naturally in the drinking water of the persons commun-
ity. Is this a case of an illness caused by the substance, or
of an innate disease whose onset can be prevented by avoiding
the substance? Vary the example, now, so that the substance
is nitrogen in the air, and ask the same question.
These examples illustrate a central feature of the notion
of an innate disease. There are commonly a host of necessary
environmental conditions for the appearance of the symptoms
of a disease. If these conditions all occur naturally or in
the normal course of events, the symptoms will be counted
as those of an innate disease. But it is often unclear whether
the occurrence of a certain necessary condition is in the normal
course of events. So it will often be unclear whether a person
is afflicted with an innate disease or is, rather, susceptible to
a (noninnate) disease.
There is much more that might be said on the topic of
innate infirmities. But it is time to take such conclusions as
we have reached and see if they can be applied in our study
of innate knowledge.

INNATE DISEASES, INNATE KNOWLEDGE, AND INNATE BELIEF

Let us begin by trading one problem for another. Ques-

tions about the nature and varieties of knowledge are as con-
troversial as any philosophers are wont to consider. But on
one point, at least, there is fair agreement: At least one sort
of knowledge is a species of belief. This is so-called proposi-
tional knowledge or knowledge that, commonly attributed by
locutions like Christopher knows that the earth is round.
8 INTRODUCTION

Not every belief, of course, is an instance of knowledge. False

beliefs are counted out; and even among true beliefs further
discrimination is needed. Specifying the principles of dis-
crimination is a problem of celebrated difficulty. Happily, it
is a problem we can conveniently avoid. For innate knowledge,
on the view of most of those who hold there is any, is innate
propositional knowledge. And if there is innate propositional
knowledge, there are innate beliefs. Let us see what sense we
can make of the doctrine that people have innate beliefs. This
will prove problem enough so that we need not feel guilty
about leaving to others the question of whether innate beliefs
are instances of innate knowledge.2
By taking innate belief in exchange for innate knowledge
we have traded up to a more manageable problem. The notion
of belief is not without its puzzles, of course. Still, every
analysis must take something as clear. So let us presume that
our workaday grasp of the concept of belief is sufficient
for the task at hand. Before attending to innate belief, it is
worth reminding ourselves that beliefs need not be objects of
current reflection. We all now believe many propositions we
are not presently thinking about, and some we have never
consciously entertained. Thus, in all likelihood, you have
long believed that your left thumb is smaller than the pyramid
of Cheops, though you have never reflected on the belief
until now. Following familiar terminology, we will call those
of our beliefs that are currently being entertained occurrent
beliefs and those we are not currently entertaining dis-
positional beliefs.
Enough said on the topic of belief. Let us ponder, now,
what might be meant by innate belief. According to
Descartes innate in innate idea has the same sense it has
in innate disease. Pursuing our strategy of following up
this hint, let us see how well our analysis of innate disease
can be adapted to innate belief. Making appropriate changes,
our account energes as follows:
A person has a belief innately at time t if, and only if, from the begin-
ning of his life to t it has been true of him that if he is or were of the
appropriate age (or at the appropriate stage of life) then he has, or in
the normal course of events would have, the belief occurrently or dis-
positionally.
 INTRODUCTION 9

This account is not without its virtues. As our introductory

quote from Descartes suggests, we can use it to wind our way
through some of the more obvious moves in the debate over
the doctrine of innate knowledge. Infants, the doctrines de-
tractors argue, believe nothing; or if they have some beliefs
they surely do not include the sophisticated propositions
proposed by the doctrines advocates. But what sense is there
to the claim that one of a mans beliefs is innate if he did
not have the belief at birth? Here our account has a ready
answer. One can have a belief innately without believing it
(occurrently or dispositionally) at birth much as one can have
a disease innately without showing its symptoms at birth.
There are, however, other problems that our Cartesian
(or dispositional) account dodges less successfully. Most
critical are problems with the interpretation of the qualifica-
tion in the normal course of events. While on the topic
of innate diseases we took note that the phrase was uncom-
fortably vague. Still, we had a passable intuitive feel for cases
that were to be clearly counted in or clearly counted out.
The trouble with the dispositional account when warped into
an analysis of innate belief is that the same intuitions seem to
swell the ranks of innate beliefs beyond all tolerable limits.
For they seem to count in just about all banal truths about
commonplace objects. In the normal course of events children
are disposed to develop the belief that night follows day and
day follows night, that things fall when dropped and that
drinking water quenches thirst. Yet surely a notion of innate-
ness distended enough to count these beliefs as innate is
bereft of philosophical interest.
In the face of this difficulty we might consider a more
liberal construal of the normal course of events, allowing in
those intuitively abnormal cases of children raised in a world
of total darkness, without gravity or water.3 Following this
strategy we would read the normal course of events as
any physically possible course of events. But this tack is
in danger of running aground on the opposite shore. If we
allow as normal circumstances that are sufficiently bizarre,
it seems likely that our account will count no beliefs as innate.
Although the issue is an empirical one, it would be surprising
10 INTRODUCTION

if it were shown that there are some beliefs people acquire

no matter how bizarre their experiences may be. Beliefs,
after all, involve concepts. One cannot believe that armadillos
are animals without having the concept of armadillo. Nor can
one believe that everything is identical with itself if one is
without the concept of identity. And, I suspect, for any
concept there is some course of (physically possible) experi-
ence which would leave a child without the concept.
Is there, perhaps, some middle course, some way to
construe the normal course of events which will leave the
dispositional concept of innate belief neither empty nor
cluttered with unwelcome occupants? One possibility is
suggested by our recent observation on the interdependence
of beliefs and concepts. Let us allow that having sufficiently
exotic experiences a person may find himself lacking any given
concept. Still, there may be beliefs that innately accompany
concepts; given that a person has had experience sufficient to
acquire the concept, he will be disposed to develop the beliefs
in the natural course of events. The beliefs, then, are condi-
tionally innate. Here, of course, we must interpret the residual
reference to the natural course of events liberally, allowing in
any experience compatible with the person having the concept.
A more restricted construal would have us again class many
banal beliefs as innate.
There is also a new danger. Having a particular concept
may entail having certain beliefs involving the concept. To
take an extreme case, it would be absurd to say a person had
the concept of an armadillo but held no true beliefs about
armadillos.4 If it is the case that having a certain concept
entails having certain specific beliefs, then the claim that these
beliefs are conditionally innate is vacuous. A belief is condi-
tionally innate if a person is disposed to acquire the belief
on acquiring a given concept. But if acquiring a concept con-
sists, in part, in acquiring the belief, then the claim that the
belief is conditionally innate amounts to the tautology that
if someone has a belief then he has it.
Despite this danger, the concept of conditional innateness
remains a plausible candidate in our quest for a philosophically
interesting unpacking of the dispositional notion of innate
 INTRODUCTION 11

belief. While some beliefs may be conditionally innate only

in the vacuous way lately considered, others may be condi-
tionally innate for nontrivial reasons. These will be those
conditionally innate beliefs the holding of which is not entailed
by having the concept they embody. Whether there be such
beliefs is open to dispute. But I see no straightforward
argument that there are none.
Still, it would be nice to have some examples. Perhaps
one of Kants examples of synthetic a priori knowledge can
be bent into an illustration. Kant held that the truths of
elementary arithmetic, like 7 + 5 = 12, were known a priori.
He also contended that the judgments these truths express are
nowhere contained within the concepts they employ. They
are synthetic not analytic truths. Now if we construe a priority
as conditional innateness and if we take the claim that 7 + 5
= 12 is synthetic to entail that having each of the concepts
involved does not entail having the belief that 7 + 5 = 12,
then the belief that 7 + 5 = 12 is a nonvacuous example of
conditional innateness. But the example is not entirely a
happy one. Quite apart from its dubious Kant scholarship, the
claim that 7 + 5=12 is synthetic is at best a matter of
controversy. Frege and the logicists who followed him under-
took to show that it was analytic. While in more recent times
Quine and others have denied that there is any distinction to
draw between analytic and synthetic truths. Here the course
of our investigation into the dispositional notion of innate
belief merges with the dispute over the nature and existence
of the analytic-synthetic distinction. To pursue Descartes
suggestion any further along the path we have come would
take us too far from the central concerns of this essay.

Our interest in conditional innateness was provoked by

the quest for some plausible way to construe the normal course
of events which would be more liberal than the construal
invoked in our concept of innate disease but more restrictive
than mere physical possibility. The proposal was that we
relativize innate beliefs to specific concepts, and allow as
normal any course of events sufficient for the person to have
the concept. This move suggests a still more permissive
12 INTRODUCTION

account of normalcy within the boundaries we have staked

out. Rather than demand normal experiences be sufficient
for the acquisition of some specific concept, we can relax
our requirement and demand of normal experience only that
it be sufficient for the acquisition of some concept or other.
Or better, we can drop the reference to concepts altogether
and take as normal any course of experience that is sufficient
for the acquisition of some belief or other. A belief is innate
for a person, then, if he is disposed to acquire it under any
circumstances sufficient for the acquisition of any belief. Here
we have a second proposal on how the notion of dispositional
innateness might be employed in an account of innate belief.
As in the case of conditional innateness, it is not obvious
that there are beliefs innate in this sense. Nor, so far as I can
see, are there straightforward arguments that there could be
none.

Before leaving the topic of dispositional innateness, let

us pause to explore one proposal of considerable interest
which is not directly in the line of our current reflections. We
have lately observed that having a concept, in one plausible
sense of this nebulous notion, may involve having certain
beliefs. But there is another sense to this notion which is
quite independent of belief. To illustrate, suppose an animal
or an infant can discriminate red from nonred things; it can
be conditioned to respond to red stimuli and can be taught
simple tasks that presuppose the ability to discriminate between
red and nonred things. We might, under these circumstances,
say that the animal or the child has the concept of red even
though it has no beliefs about red things. Concepts in this
sense are prime candidates for innate concepts in the sense of
innateness modeled after innate diseases. For if simple con-
ditioned learning is to take place, the organism that does the
learning must be able to discriminate stimuli that are being
reinforced from those not reinforced. And since most organisms
can, in fact, be conditioned to some stimuli from birth, some
concepts must be innate.5 This is the theme elaborated by
W. V. Quine in his contribution to this volume.
 INTRODUCTION 13

It is time to take stock of our progress so far. Our strategy

was to follow up Descartes suggestion by seeking an account
of innate belief on the analogy of our analysis of innate disease.
We discovered that the analogy is not so straightforward as
Descartes may have thought. For buried in the notion of
innate disease is an appeal to the normal or natural course
of events. And while our intuitions about what is normal or
natural serve passably well when we attend to innate disease,
the same intuitions yield an intolerably broad notion of innate
belief. In casting about for a more restrictive account of what
is to be allowed as normal, we have come upon two possi-
bilities. The first led to the concept of conditional innateness;
the second counted a course of experience as normal if it
led to the acquisition of any belief at all. These alternatives
are at best tentative proposals. There is much work yet to be
done on the dispositional account of innateness. But in this
essay we must abandon the topic here and turn our attention
to a quite different attempt at explicating the notion of
innate belief.
THE INPUT-OUTPUT MODEL:
ANOTHER APPROACH TO INNATE BELIEF

The dispositional account of innateness was suggested by

Descartes analogy between innate ideas and innate diseases.
The alternative account that is our current topic can be coaxed
from the exchange between Socrates and the slave boy in
Platos Meno. Though Socrates succeeds in eliciting from the
boy the solution to the problem he has posed, Socrates none-
theless insists that he has not taught the boy anything. Rather,
he tells Meno, he has uncovered something that was in the boy
all along. Thus Socrates claims that the boy has some sort of
innate belief. But it is clear that it is not a dispositionally
innate belief. For at the beginning of the interrogation the
boy does not believe what he later recollects, nor need he
ever have come to believe it. The questioning played a crucial
role. There is no suggestion that the belief would have arisen
without the questioning as part of the normal course of
events. Moreover, the questioning did not serve to supply the
boy with new concepts. He seems to have all the requisite
14 INTRODUCTION

conceptual apparatus before the questioning begins. So the

beliefs he comes to hold are not conditionally innate. How
are we to understand this nondispositional sense of innateness
Plato seems to be using?
One idea that takes its cue from Platos remarks is to
view the role of the Socratic interrogation as akin to the role
of a trigger or a catalyst. It sets off a process that results in
the acquisition of the belief. But, as a catalyst is not part of
the end product of a chemical reaction, so the questioning
process does not supply the content of the belief. The content
of the belief was contained within the boy much as the content
of a tape recorded message was contained upon the tape. The
questioning experience, like the throwing of the tape recorders
switch, serves only to set off the appropriate mechanism. On
this model we can begin to make sense of the claim that the
beliefs contained within the boy are innate even though they
require certain sorts of experiences to bring them out.
There is at best scant textual evidence for the hypothesis
that Plato would have expanded his doctrine along the lines
that we have taken. With later authors, however, it is quite
clear that they flirted with the model we are considering.
Leibniz, for example, contends the mind has a disposition (as
much active as passive) to draw [necessary truths] from its
depths; although the senses are necessary to give it the occasion
and attention for this and to carry it to some rather than
others.6 He makes much the same point with his favorite
metaphor. It is a disposition, an aptitude, a preformation
which determines our soul, and which brings it about that
[necessary truths] may be derived from it. Just as there is
a difference between the figures which are given to the stone
or marble indifferently, and those which its veins already mark
out, or are disposed to mark out, if the workman profits by
them.7 A natural reading of the metaphor is that in acquiring
knowledge of necessary truths the mind uses experience only
as a catalyst providing the occasion or cause for the knowledge
being uncovered, but providing little or none of the content
of the knowledge, just as when an appropriately grained block
of marble is transformed into a statue the workman need only
tap and chip a bit to uncover the figure. In the selection from
 INTRODUCTION 15

Cartesian Linguistics included in this volume, Noam Chomsky

finds evidence of this view of experience as a trigger for innate
cognitive mechanisms in thinkers as diverse as Schlegel and
Herbert of Cherbury.
In several of his discussions of the catalyst or trigger
metaphor, Chomsky suggests a variant on the figure. He
proposes that we look on belief acquisition as an input-output
process, with sensory experience as input and belief as output.
If the beliefs that result from a particular pattern of sensory
experience are richer or contain more information than the
experience, then this added information must be the minds
contribution. If the total sensory input up to a given moment
in time is poorer in information than the beliefs acquired to
that moment, the excess information is innate. Where the
disparity is particularly great, the sensory input contributes
little or nothing to the belief acquired. It acts merely as a
trigger, setting off the innate cognitive mechanisms.
It is important to see that though Chomskys suggestion
is couched in terms rather more modern than those used by
Leibniz, it is nonetheless little more than a metaphor. Chomsky
is proposing that belief acquisition be viewed as an input-output
process and that the mind is interestingly similar to an input-
output device. If we are to pursue this proposal seriously,
trying to turn it into more than a suggestive metaphor, we
should have to give some account of how we measure the
comparative richnessor information contentof experiences
and beliefs. Existing accounts of information content will not
do. They treat of the information in a proposition or sentence,
not the information in a belief or stretch of experience. Also,
familiar accounts of information content count logical truths
as containing minimal information. So adopting such an
account for our present purposes would lead us to exclude
belief in logical truths as innate beliefs, though such beliefs
have often been taken as paradigms of what is known innately.
Even without any developed account of the appropriate
notion of information content, we may note one quite funda-
mental difference between the input-output model of innateness
and the pair of dispositional concepts developed previously.
On either dispositional account the hypothesis that there are
16 INTRODUCTION

innate beliefs is moot. On the input-output model, however,

there can hardly be any doubt that many beliefs are in part
innate. Most any empirical belief, for example, will be richer
in information content than the experience that led to its
acquisitionand this on any plausible account of the appro-
priate information measures. This is a consequence of the
philosophical commonplace that the evidence a person has for
an empirical belief rarely entails the belief. While we may
come to believe that all armadillos are omnivorous by observ-
ing the eating habits of a fair sample of armadillos, the general-
ization is not implied by any number of propositions attributing
varied tastes to particular armadillos.8 In the case of mathe-
matical or logical beliefs it is rather harder to specify the
relevant experiental input. But again it seems that on any
appropriate measure of information content the information
contained within our mathematical and logical beliefs outruns
that contained in our total sensory history.
The upshot of these observations is that when pursuing
the input-output model of innate belief, the interesting question
is no longer whether there are beliefs that are (in part) innate.
Rather what is interesting is to what degree our various beliefs
are innate. Also of interest is the detailed story about the
cognitive mechanisms that lead from sensory input to belief.
In developing his theory of the acquisition of language, Chomsky
is making a tentative effort at sketching in some of these
details.

A PRIORI KNOWLEDGE AND INNATE IDEAS:

TWO MORE THREADS TO UNTANGLE

The project with which we began this essay was to un-

tangle some of the strands that run through the long history
of the argument over innate ideas and innate knowledge. So
far we have succeeded in separating out two basic concepts
which historically have often been run together. The several
related notions of innateness which flow from Descartes analogy
between innate ideas and innate diseases contrast sharply
with innateness conceived on the input-output model. It
would be tempting to see much of the historical debate over
 INTRODUCTION 17

the innateness doctrine as a consequence of the failure to

distinguish these two sorts of innateness. But, though tempt-
ing, it would be inaccurate. For, though some of the historical
(and modern!) debates can no doubt be traced to the failure
to distinguish these two concepts, problems were multiplied
by still other confusions. In particular there is a pair of
notions whose history is wound together with the history of
the idea of innateness. One of them is the concept of a
priori knowledge.
In the preceding two sections we retreated from tackling
the notion of innate knowledge, and focused instead on innate
belief. In so doing we avoided need to talk of warrant or
justificationa. property a true belief must have to be an in-
stance of knowledge. We thus avoided confronting the issue
of a priori knowledge, which is tied to the concept of justifi-
cation.
For some of the propositions we know our justification
is (at least in part) to be traced to sensory experience. But,
on the view of many philosophers, we know some propositions
whose justification is entirely independent of experience. These
are the propositions we know a priori. Our belief in these
propositions may have been (in part) caused by experience.
But the justification we have that makes instances of a priori
knowledge more than mere belief is not to be found in the
experience that caused the beliefs, nor in any other experience.
To say that a bit of knowledge is a priori, then, is to say
something about its justification, while to say that a belief is
innate is to say something about its cause or genesis.9
Though the distinction between innateness and a priority
seems passably clear, the two have not always been distinguished.
Thus Leibniz writes: . . . very often the consideration of the
nature of things is nothing else than the knowledge of the
nature of our mind and of those innate ideas which we do not
need to seek outside. Thus I call innate those truths which
need only this consideration in order to be verified.10 And
elsewhere: . . . it is always clear in all the states of the soul
that necessary truths are innate and are proved by what is
internal.11 Here there is maddening tangle. If the truths
are verified from within, proved by what is internal, then it
18 INTRODUCTION

is their justification that is independent of experience. So it

is a priority not innateness that is at issue. Perhaps Leibniz
thought that all and only innate knowledge was known a
priori. But once the two have been distinguished the claim
that they coincide in extension is itself in need of justification.

The second of the pair of concepts whose history is

bound up with the history of the doctrine of innate knowledge
is the notion of an innate idea. Talk of belief or knowledge
slips easily into talk about ideas. Indeed, in previous pages I
have occasionally slid back and forth from one to the other
with a studied equivocation. But though sometimes talk about
ideas is but a colloquial variant on talk about knowledge, it
is not always so. For the Classical Rationalists, who have
loomed large in our discussion of innateness, had quite a
unique use for the term innate idea. Their doctrine of
innate ideas is to be understood against the background of the
Aristotelian scholasticism that flourished in the late Middle
Ages, and it admits of no tidy summary. Happily we need not
here attempt an explication of their views, since the job is
done with great clarity in Robert Adams contribution to
this volume.12
While the controversy between Classical Rationalists and
Classical Empiricists on the topic of innateness was focused
as often on innate ideas as on innate knowledge, the modern
rationalists, who would argue their innateness doctrine
from the theories of modern grammar, talk mostly of innate
knowledge. So in the remaining section of this essay, I will
attempt a brief sketch of the nature of modern grammar. I
can then indicate where questions of knowledgeinnate or
otherwiseare likely to arise.
GRAMMAR AND KNOWLEDGE13

A grammar is a theory. The grammarians principal data

are the judgments speakers make about expressionsjudgments,
for example, that expressions are or are not grammatical
sentences, that sentences are ambiguous, that pairs of sen-
tences are related as active and passive or as simple declarative
 INTRODUCTION 19

and yes-no (or wh-) questions, 14 and a host of others. Roughly

speaking, the grammarian tries to build a theory which will
entail that expressions have the properties speakers judge them
to have. If a grammar is to be an adequate theory of the
language of a speaker, it must entail that an expression has
a given grammatical property if the speaker would judge the
expression to have the property.
This brief account must be modified in several directions.
First, speakers judgments are not the only data a grammarian
may use. Data about what a speaker does and does not say
in unreflective speech, data about pronunciation peculiarities
and a host of other phenomena may also be taken into ac-
count. Also, a grammar is an idealized theory. The grammarian
will systematically ignore certain discrepancies between what
his theory says of some expressions and what the speaker says
of the same expressions much as, in the theory of ideal gases,
we systematically ignore deviations between predicted corre-
lations of temperature, pressure and volume and observed
correlations. In both cases the motive is much the samethe
expectation that construction of a complete theory that
accurately describes all the phenomena is best approached by
breaking the job into several parts, first giving the idealized
theory, then explaining the deviations.
Commonly a grammar will consist of a set of rules (phrase
structure rules, transformational rules, and perhaps some others)
and a set of definitions. 15 The definitions and rules entail a
variety of statements. They entail many of the form:
S is a grammatical sentence
where S is replaced by the name of an expression; many of
the form:
e is the subject of sentence S;
many of the form:
Sentences S and S are related as active and passive, etc.
It is these consequences of the rules and definitions which
must agree with speakers judgments. The rules and definitions
20 INTRODUCTION

form an integrated empirical theory, and both rules and

definitions may be modified in the face of recalcitrant
data.
The grammarians theory construction does not stop with
a grammar. Having made some progress at grammars for several
languages, he turns his attention to linguistic theory or the
theory of grammars. Here the goal is to discover linguistic
universals, general features of the grammars of human languages.
These universals may be general constraints on the form of
grammarsthat all are divided into phrase structure and trans-
formational components, say, or that all use rules only of a
specified sort. The universals may also include particular
rules or definitions which are the same in the grammar of
every natural language. If any rules or definitions are universal,
they need no longer be specified along with the more idio-
syncratic details of individual grammars.
The linguistic theory is also concerned with the acquisition
of grammarhow a person comes to have the grammar he does.
Here the strategy is to find a function ranking humanly possi-
ble grammars. The goal is to find a function that ranks
highest among humanly possible grammars that grammar which
the child actually acquires, when we first exclude from the
class of humanly possible grammars all those that are incompa-
tible with the observed utterances and other data available to
the child. Specification of linguistic universals and a measure
function of the sort described would provide a (low level)
explanation of how the speakers of a language come to have
the grammar they do.
These are the two sorts of theories the grammarian con-
structs. If the theories are correct, they will describe certain
facts about speakers linguistic intuitions (for grammars) and
certain facts about all human grammars (for linguistic theory).
About what aspects of these theories might speakers be thought
to have knowledge? In the essays that follow there are, I
think, three distinct proposals.
First, it might be thought that speakers know the linguis-
tic universals, that they know (perhaps innately) that all human
languages have phrase structure and transformational rules,
or that the grammar of every language contains some specific
 INTRODUCTION 21

rule or that in every natural language an expression is a noun

phrase if and only if . In short, this first suggestion is
that speakers know that p where p may be replaced by any
statement belonging to linguistic theory.
Next it might be held speakers know that the particular
rules of the grammar of their language are rules of the grammar
of their language, or that they know the definitions that,
along with the rules, constitute the grammar of their language.
Third, and most plausibly, it might be thought that speak-
ers have knowledge of the consequence of the rules and defini-
tions of their grammar. If this suggestion is correct then
speakers of English will know that Mary had a little lamb is
grammatical, that Mary is its subject, and that it is several
ways ambiguous. Each of these views is considered in the
debates that follow among Chomsky, Putnam, Katz, and
Harman.

NOTES

1. Cf. Nelson Goodman, Fact, Fiction and Forecast (2d ed.),

chap. II, sect. 2 (Indianapolis: Bobbs-Merrill, 1965).
2. Some argument on this topic can be found in the debate between
Professors Hart and Goldman included in this volume. Cf. also R. Edgely,
Innate Ideas, in Knowledge and Necessity, Royal Institute of Philosophy
Lectures, vol. 3 (London: Macmillan, 1970).
3. Note that each of these circumstances must be counted as
abnormal by the account of normalcy required for innate diseases.
Symptoms that appear in the absence of gravity are not symptoms of an
innate disease.
4. This is, of course, not sufficient to establish the stronger claim
that there is some specific belief the holding of which is necessary for
the possession of the concept.
5. Though this argument seems straightforward enough, there is a
problem buried here. Let me indicate it briefly. We have contended
that conditionability requires a preexisting concept or, as Quine would
have it, a quality space or qualitative spacing of stimulations. Now
in the case of colors, tones, and other relatively elementary sensory
qualities, our contention seems to have some rudimentary explanatory
value. We would like to know much more about quality spaces. But
still, to say an organism prior to conditioning must have a qualitative
spacing of stimulations seems to add something to the bare observation
that the organism is conditionable. Now contrast these cases with other
22 INTRODUCTION

instances of conditionability. Some organisms (some people, for example)

can be conditioned to respond differently to paintings in the style of
Rubens, as contrasted with paintings in the style of Monet. Other organ-
isms (I presume) cannot be so conditioned. The case seems, for all
we have said, quite analogous to the case of colors and tones. But here
it seems perverse to postulate that the conditionable organisms have a
preexisting quality space. Such a move appears explanatorily vacuous.
It adds nothing to the bare observation that the organisms are condition-
able. All this is impressionistic. But if my impressions are correct we
are left with a problem: Why is the postulation of a preexisting quality
space plausible in one sort of case and perverse in the other?
6. Leibniz, New Essays (1703-1705), I, i, 5.
7. Ibid., 11.
8. As this observation indicates, the notion of innateness built on
the input-output model is not inimical to empiricism. Humes doctrine
of natural belief required an inborn faculty or mental mechanism by
which we acquire our beliefs about matters of fact. The beliefs acquired
are not entailed by the sensory evidence we have for them. Thus, in the
sense of innateness under consideration, they are in part innate. This
theme is taken up briefly in Professor Harmans essay included in this
volume.
9. There is some precedent for a rather narrower notion of a priority.
Kant, for example, made it a criterion for a judgment being a priori
that it be thought as necessary. Following Kants lead we might say
that a person has a priori knowledge of a proposition (in this narrower
sense) when he knows the proposition and his justification is independent
of experience and the proposition is necessarily true.
10. Leibniz, New Essays, I, i, 21; emphasis added.
11. Ibid., p. 5; emphasis added.
12. The reader should be warned that not each occurrence of innate
idea and kindred expressions in Descartes or Leibniz is intended in the
sense Professor Adams explains. Sometimes their use of the term is better
taken as meaning innate knowledge or a priori knowledge.
13. This section is adapted from my essay, What Every Speaker
Knows, Philosophical Review, Vol. LXXX, no. 4 (October, 1971).
14. E.g., Max went to the store and Did Max go to the store? are
related as simple declarative and yes-no question; Max went to the store
and Who went to the store? are related as simple declarative and wh-
question, as are Max went to the store and Where did Max go?
15. The literature on modern or generative grammars is vast and grow-
ing. A good starting place for the reader new to the subject would be N.
Chomsky, Aspects of the Theory of Syntax (Cambridge, Mass.: MIT Press,
1965).